Guide or Tutorial for how to evaluate subs?

[Lightpath]

Hey all-

There's so much wonderful information on this site, it's like drinking from a fire hose!

I was wondering if anyone had a step-by-step tutorial/guide on how to evaluate subframes.  I would like to figure out what my optimal exposure is for my setup.  I have followed many of those tutorials and used the excellent tools in PI but the results were only semi-conclusive.  I would like to be able to measure my subs and actually understand what makes a good sub-  That would help me tune my exposure time, and help me understand what I should keep and what I should throw away.

I've used the excellent subframe selector script and the amazing new subframe selector process, but I still find I don't quite know what I'm actually looking for in a good sub.  I would like to be able to thoughtfully create my own formulae for rejecting/accepting frames, rather than blindly copying those that I see around this forum.

My approach to Astrophotography so far has been to take one problem at a time-  For example, getting my focus working well, getting tracking working well, getting flat frames properly exposed, and now I'm up to processing, since I think I have my mechanical bugs (mostly) worked out, and I'm trying to take it one step at a time.

Thanks very much!

Mark.

[Lightpath]

Wow, no one has any ideas? 

It occurs to me that getting the subs right is probably one of the most important parts of this process!

[pfile]

well... if you look at some of the stuff david ault has put out there regarding SFS approval expressions, or maybe jon rista, i guess it's all kind of mom and apple pie stuff that we can all agree on - good subs have low eccentricity, high star counts, low FWHM, etc.

have you dissected one of those expressions?

rob

[Lightpath]

I have, actually, I generally understand what they're doing, since PI is so awesome about variable naming, but getting an idea of "why" the FWHM should be within a certain range, as an example, is what I'm kind of looking for.  I generally understand what's being done in the equations, but I'd like to be able to say to myself "Ok, I know I want the FWHM to be in [foo] range because [bar].  I was just wondering if someone had written something up sort of like "This is how to make sure you're getting a good subframe" guide.

One of the things I've been struggling with is using the image evaluation tools to get my exposure length dialed in for example.  Yes you can use subframes that aren't taken with the optimal exposure, but it'd be better to just make sure you have excellent subs from the start, and tweak your system as much as possible to get the best data you can get.

Maybe I just need to keep digging and reading, or perhaps I'm approaching this incorrectly.

Thanks for the reply!

[ngc1535]

I might say there are different ways of thinking about evaluating subframes. For example, for some data S/N is more important than resolution. So throwing out frames based on a strict range of PSF/FWHM doesn't make sense. In fact, as many people know that have watched me deliver presentations, I tend to find every excuse in the book to *keep* data rather than cull it. Too often, I feel, image processors throw out data without taking into account the proper usage of weighting and rejection. In fact the weighting given to a subframe is probably (based on signal or noise) more important than any of the other "beauty" factors (PSF, eccentricity...).

Say you had a set of 15 subframes. You are now well positioned to combine them with good rejection choices. If you only have 6 subframes, your choices are more limited and the calculus of what makes a good subframe changes. But back to the 15 frames... if you have 15 frames but 4 of them show wind shake or bad guiding (loss of guide star). Most people throw those frames away based on the subframe selection formula. In my opinion that is too draconian. Instead choose the right rejection method and parameters and use the frames with the proper weighting and gain the benefit of the added S/N improvement.

Just something to think about. I think the reason you aren't seeing that many answers is that there are a few nuances and variations to consider.


-adam

[pfile]

well i guess the FWHM is going to be a function of your seeing and the aperture of the telescope (and what filter you are using). i suppose the only way to know what is a good FHWM and what is not a good FHWM for you is to compare a whole bunch of subs taken thru the same filter on a variety of nights. the only problem with this empirical method is that it may be impossible to reach the theoretical capability of your system if your seeing is consistently poor. but in some sense it doesn't matter - you have to work with the sky you have.

here's a thread from some variable star observer guys which is pretty interesting:

https://www.aavso.org/fwhm-expectations-larger-aperture-instruments

adam of course has a point - i know that rick has mentioned just throwing everything he has at one integration to try to boost the SNR of the background and faint areas, and then more carefully culling for an integration of the target itself. of course that then leaves you with the problem of trying to merge the data, but i guess the point is sometimes it might make sense to integrate a bunch of 'crap' data.

rob

[chris.bailey]

I suspect there are numerous equally valid approaches. I just cull the obvious outliers in terms of eccentricity/FWHM i.e anything much more than 2 sigma away from the mean. I always base weights on noise evaluation in image integration as I have never seen much benefit from using complex expressions in the Sub Frame Selector.

[Lightpath]

I might say there are different ways of thinking about evaluating subframes. For example, for some data S/N is more important than resolution. So throwing out frames based on a strict range of PSF/FWHM doesn't make sense. In fact, as many people know that have watched me deliver presentations, I tend to find every excuse in the book to *keep* data rather than cull it. Too often, I feel, image processors throw out data without taking into account the proper usage of weighting and rejection. In fact the weighting given to a subframe is probably (based on signal or noise) more important than any of the other "beauty" factors (PSF, eccentricity...).

Say you had a set of 15 subframes. You are now well positioned to combine them with good rejection choices. If you only have 6 subframes, your choices are more limited and the calculus of what makes a good subframe changes. But back to the 15 frames... if you have 15 frames but 4 of them show wind shake or bad guiding (loss of guide star). Most people throw those frames away based on the subframe selection formula. In my opinion that is too draconian. Instead choose the right rejection method and parameters and use the frames with the proper weighting and gain the benefit of the added S/N improvement.

Just something to think about. I think the reason you aren't seeing that many answers is that there are a few nuances and variations to consider.


-adam

THIS is exactly the type of information I'm looking for.

This demonstrates a deeper knowledge of how to adjust both evaluating frames and stacking parameters to make the best use of gathered photons as possible.  How did you learn/figure out that knowledge?  I find it difficult to move on to higher level issues (removing background noise, colour balance, etc) without actually closing the loop on my imaging system, and making sure I'm getting the best data I can. 

Then I'd like to be able to take the data that doesn't quite make the grade, and salvage what I can.

I think you're right-  This is a very nuanced topic, but I think perhaps it's a topic that we don't discuss enough?

I am typically dealing with hundreds of subframes.  I have a tiny, low patch of sky to image in, and tend to image the same thing for many nights in a row in order to get lots of data.  I'd like to be able to evaluate and save as much of that data as possible.

Do you have a link to one of your presentations?

Thanks very much for your thoughts!

edit:  I found this page extremely useful:
http://www.lightvortexastronomy.com/measuring-your-camera-sensor-parameters-automatically-with-pixinsight.html

[ngc1535]

Hi LightPath,

There exist many presentations I have given that are available on-line and speak to this topic in a way that is independent of software. When coupled with a program like PixInsight of course, you leave me no choice but to plug myself (please see AdamBlockStudios.com). Ironically one the sections I would like to complete is exactly the one you want.
I am waiting for a response from Juan to make certain I get it right before moving forward. (Juan, if you are back...I am really looking forward to hear from you!) But in a more general way, I demonstrate the thought processes I indicated above in my workflow sections.

As you mention in your comment, the evaluation of subframes is both a window into acquisition issues as well as the ways in which the sky changes and affects data. An understanding of these things is, of course, important. For example, if you simply image a target for long periods as it is setting- due to atmospheric extinction and increasing airmass you will clearly see a decline in quality (with everything else being equal) in the subframes. So you should look for the weights of images to show this decline as well. This is the kind of "deeper" knowledge I use when looking at data. Another example is having a firm understanding of rejection algorithms and behaviors. The point is, can you look at your set of images (say blink them) and know a priori if things you see (cosmic rays, satellite trails, bad guiding blips...etc) will be rejected cleanly? This is a useful skill to have when evaluating subframes. (To go one step further on this, this initial look at subframes will also inform whether or not you need to use LocalNormalization.)

I know I am not spelling everything out- it would be a small essay- but again I hope I am giving you some things to consider.

-adam

[Lightpath]

Thanks Adam, I very much appreciate your insight and information.

ATM I'm not in a financial position to subscribe to your videos, but I think it's great that they are available!

Thanks-